Sensor scheduling for bearings-only tracking with a single sensor

In this paper, two types of optimality criteria for scheduling a single passive sensor in bearings-only tracking are investigated. Firstly, we examine the case of optimising sensor trajectory under a classical control frame work where the optimality criteria are based on Fisher information matrix. A first order numerical algorithm is discussed. Secondly, the sensor scheduling problem is treated as a partially observed Markov decision process where information divergence criterion is ultilised as a reward function. A grid based search strategy is adopted to find feasible sensor courses which can maximise the reward between prior and posterior densities of target state. Technically, accumulated expected reward is evaluated at each grid point to avoid using measurements. The best sensor trajectory is found by selecting those sensor locations of maximum accumulated expected reward in the time period of interest. The performance of these two techniques are compared via computer simulation in a typical single sensor bearings-only tracking scenario used in the literature. Comparison results are discussed accordingly.